Introduction to AI in Learning: Designing the Future

Abstract

The introduction chapter focuses on the main questions of the whole book AI in Learning: Designing the Future: (1) How is learning changing when human learning and machine learning are connected and what consequences does this conjunction have for education, also for working life as lifelong learning and (2) what kind of ethical issues are emerging with AI in education from the viewpoints of schools and other learning environments. The chapter first summarizes how recent AI technologies provide several options for learning and educational services and how AI is applied already in societies. In education and learning, many advanced techniques are already available, and we have tentatively promising findings. However, the accelerating pace of development of technology expands AI’s potentialities in education, so we need extensive new research about educational implementations and their effects on human learning and people’s lives. The chapter also summarizes how different chapters provide new research on AI in learning and education.

1 The Aim and Background of the Book

Artificial intelligence (AI) is changing the world radically. It impacts societies, organizations, work, and education, and it is becoming more and more part of everyday life. The surge of AI requires analysis and foresight to determine what it may mean in education and for learning. This book is based on contemporary research with Artificial Intelligence in educational settings (AIED) in educational settings. The major questions are: (1) How is learning changing when human learning and machine learning are connected and what consequences does this conjunction have for education, also for working life as lifelong learning and (2) what kind of ethical issues are emerging with AI in education from the viewpoints of schools and other learning environments. The core aim is to discover how AI-based intelligent tools and environments can augment and support human learning. In this volume, over 60 researchers in universities in China, USA, and Finland have introduced their recent research concerning how they see the potentialities of AI for education and learning. Many authors provide evidence of new applications and consequences. Many chapters also provide reflections on the newest trends in AI development and what kinds of changes they may require in adaptations by schools and working life contexts.

Our authors leap forward to share the ways AI may contribute to redesigning our future when it is applied in education and learning. This introduction has two tasks. It first draws a general picture of the state of the art of AI’s role globally and summarizes how education is a fundamental part of these processes. Secondly, the introduction summarizes the contributions of chapters. The book has four parts, each of them giving a special viewpoint to AI with meanings, relevance, and challenges when applying AI in learning and education.

2 AI in a Global World: State of the Art

The definition of AI has been in discussion since its origins. Starting in the 1950s, the core idea of most definitions has been that a machine can be intelligent because it embodies some performance elements that human brains enact such that computer systems can perform tasks that normally require human intelligence (e.g., Stone 2016; Roschelle et al. 2020; UNESCO 2021a). Based on huge developments in technology and computing sciences, we can see that AI has become a more and more complex, cross-subject and cross-disciplinary, multipurpose, global endeavor, and it is in an ongoing development process. The intelligent features have increased with advanced computer programming, for example, through neural networks in deep learning. Still many researchers remark that it is still a long way from achieving the flexibility, width of task performance, and progress in competences to reflect on and give reasons for decisions made that are typical qualities of the human mind. Nonetheless, new technologies have made AI useful for industry and business, health and medicine, transportation, and logistics as well as in many service sectors. AI has brought additional value to design, manufacturing, and products, robotics, chatbots, and automatic mobile device log in and face recognition as typical examples. All the same, researchers still observe that we are only in the spring of AI applications and much more research and development will be needed to achieve the full potential of AI for all the seasons (Stone 2016).

At the policy-making level concerning AI and human affairs, the last 5 years have demonstrated exponential growth. In China, the USA, and the European Union, many strategic plans have been published since the middle of the last decade. The recent trends can be summarized:

• In China, a discussion paper from the McKinsey Global Institute (2017), originally presented at the 2017 China Development Forum, explored AI’s potential to fuel China’s productivity and growth – and to disrupt the nation’s workforce.

• In the USA, the National AI Initiative Act of 2020 (House of Representatives 2020) became law on January 1, 2021, providing for a coordinated program across the entire federal government to accelerate AI research and application for the nation’s economic prosperity and national security.

• The European Commission’s (2020) white paper on AI sets strategic plans for how European countries will use AI in different sectors of society. AI should work for people and be a force for good in society. The European union has a strong emphasis on ethical issues (Europen Union 2020). AI is recognized for the likelihood that it will change the whole society, and the European documents correspondingly emphasize, in addition to the topic of work disruption due to job automation, key concerns for advancing trustworthy AI and its ethical requirements.

The Organization for Economic Co-operation and Development (OECD) launched 2020 “AI Policy Observatory” (OECD 2019). It tracks policy areas where AI is driving changes in the workforce, transportation, and healthcare sectors. It follows up trends and AI data use and provides a forum for national AI policies and global initiatives of different stakeholders including business, academia, and civil society. The OECD highlights that its observatory project aims to help countries to encourage, nurture, and monitor the responsible development of trustworthy AI systems for the benefit of society. OECD AI principles also recommend governments and the private sector combine their investments for research, including interdisciplinary efforts and development of AI. The future emphasis is that innovations should focus on challenging technical issues and on AI-related social, legal, and ethical implications and policy issues.

In addition to the policy-level strategies, AI is also seen as a tool for sustainable development. In April 2021, the United Nations (UN) published its Resource Guide on Artificial Intelligence AI Strategies (UN 2021). It introduces how AI can provide resources to achieve sustainable development goals (SDGs) that are related to big challenges such as climate change, hunger, poverty, inequalities, and other severe global threats. The volume has collected existing AI-based resources as well as examples from policies, strategic plans, and ethical guidelines of governments, private sectors, and other stakeholders. It also warns that AI will have unanticipated consequences that will exacerbate inequalities and negatively impact individuals, societies, economies, and the environment.

UNESCO, as a United Nations agency, has a special mandate for education and culture. UNESCO published its 2021 AI guidelines for policymakers, introducing and reviewing AI technologies in educational and their ethical challenges (UNESCO 2021a). In November 2021, UNESCO launched a report Reimagining our futures together: a new social contract for education (UNESCO 2021b). The report provides a strong appeal for the importance of education and the strategic goals in education (SDG 4) that is one of strategic goals of UN. The report has a strong message: Quality education and access to learning must be guaranteed for everyone and throughout the life course. We need a new social contract to develop education globally. Access to school alone is not enough. Currently, the biggest problem is the quality of education; what and how to learn in schools. Inequality in quality education is growing exponentially and sustainable development is also based on education. For AI, UNESCO has a double message (Niemi 2020). On the one hand, we need new technology that helps to increase access to education and increase the quality of education. And on the other hand, AI must not increase the digital gap and deepen inequities in education. Based on over 2 years of joint and wide cooperation with its member states, the recommendations for AI policy were adopted by UNESCO’s General Conference on 24 November 2021 (UNESCO 2021c). The consensus reaffirms a humanistic approach to the use of AI with a view towards protecting human rights and preparing all people with the appropriate values and skills needed for effective human–machine collaboration in life, learning, and work, and for sustainable development. It advocates for human-controlled and human-centered AI development, where the deployment of AI should be in the service of people and to enhance human capacities. It recommends that the impact of AI on people and society should be monitored and evaluated throughout value chains. The key principles emphasize that digital technologies should aim to support—and not replace—schools. We should leverage digital tools to enhance student creativity and communication. When AI and digital algorithms are brought into schools, we must be vigilant to ensure that they do not simply reproduce existing stereotypes and systems of exclusion.

3 AI in Education

AI is part of fundamental global changes and its power is increasing. Most policy-level strategic plans draw a picture at the global or whole societal level. References to education comes mainly from a perspective of changes in work and new competences needed in working life. Otherwise, education and learning are rather invisible in the policy-level documents. This concerns also ethical principles of AI that set general guidelines for trustworthy AI. Only UNESCO’s guidelines and ethical principles have focused directly on education.

However, we have reviews and ongoing research how AI has been implemented in education and learning (Bransford et al. 2006; Niemi 2021). AI has already entered education and schools in different forms. Learning Sciences has published for decades how learning analytics can help to recognize and facilitate learning processes with intelligent tools (Baker and Inventado 2014; Fischer et al. 2020; Niemi et al. 2018). Chen et al. (2020) reviewed research on AI published in education in high-quality international journals between 2009 and 2019. The review provides evidence that AI has been extensively adopted and used in administration, instruction, and learning. In administration, AI applications such as reviewing and grading students’ assignments were seen as very useful and, in some cases, even more accurate than human-based assessments. Important implementations were also applications for teachers which help them improve instruction with more knowledge about students’ learning and with interactive tools for learners’ knowledge construction and sharing. For students’ learning, AI could help them by tutoring and personalization. New technological systems have leveraged machine learning and adaptability, and curriculum and contents can be customized and personalized in line with students’ needs. Reviews and analyses of current state of the art (Chen et al. 2020; Stone et al. 2016; Timms 2016; Roschelle et al. 2020) also reveal that a transformation has happened from computer and computer-related technologies to web-based and online intelligent education systems. Often with the use of embedded computer systems but also together with other technologies, we also note the use of humanoid robots and web-based chatbots to perform instructors’ duties and functions independently or jointly with instructors. AI-related themes, such as teaching robots, intelligent tutoring systems (ITS), online learning, and learning analytics, have become common over the past several years. In many studies, big data, learning analytics, and data mining techniques have become major tools for personalized learning.

Recent AI technologies provide several options for learning and educational services which can be summarized (UNESCO 2021a; Roschelle et al. 2020):

• Natural language processing (NLP) for the use of AI to automatically interpret texts, including semantic analysis, used in translations, and for generating texts of learning contents, and supporting personalization processes.

• Speech recognition covers the application of NLP to spoken words, including smartphones, and provides AI personal assistants within games and intelligent tutoring systems, and for conversational bots in learning platforms.

• Image recognition and processing employs AI for facial recognition (e.g., for electronic documents and processes in classroom situations), handwriting recognition, text analysis (e.g., to detect plagiarism), image manipulation (e.g., for recognizing deepfakes), and for autonomous scoring and grading.

• Autonomous agents use AI in computer game avatars, software bots, virtual learning spaces, smart robots.

• Affect detection employs AI to analyze sentiment in text, behavior, and faces.

• AI underlies data mining algorithms for predictive learning diagnoses, progress forecasting, socio-emotional well-being analysis, financial predictions, and fraud detection.

• Artificial creativity uses AI in systems that can create new kinds and exemplars of photographs, music, artwork, or stories.

In the last 10 years, AI has taken big steps in education and learning with a new method of computing and advanced technology for using and integrating multimodal data. The multisector expert group (Roschelle et al. 2020) convened by the nonprofit organization Digital Promise drafted scenarios for how AI will influence education. They foresee that AI-based learning goes far beyond what was earlier possible with tracing users’ learning paths through keyboard strokes or eye movements in learning analytics. The advanced human-machine interface provides AI-related functions including natural language interaction, speech recognition, and detecting learners’ emotions. AI allows sensing, recognizing patterns, representing knowledge, making and acting on plans, and supporting naturalistic interactions with people and support learners with varied strengths and needs, allowing students to use handwriting, gestures, or speech as input in addition to more traditional keyboard and pointer input. The expert group also sees that AI can support learning in terms of orchestrating complex learning activities with multiple people and resources, augmenting human abilities in learning contexts, expanding naturalistic interactions among learners and with artificial agents. It broadens the competencies that can be assessed and reveals learning connections that are not easily visible. These approaches go beyond familiar design concepts for individualized, personalized, or adaptive learning. All these new opportunities bring many ethical challenges and these should be urgently investigated.

As a conclusion of our brief survey of the current state of art in AI for education and learning, we can see that AI is massively applied already in societies and globally. In education and learning, many advanced techniques are already available, and we have tentatively promising findings (e.g., Niemi 2021). However, the accelerating pace of development of technology expands AI’s potentialities in education, so we need extensive new research about educational implementations and their effects on human learning and people’s lives. The more AI is applied in education and learning, the more we need reflections on and solid grounds for ethical use of AI.

4 The Structure and Contents of the Book

The book is based on the most recent research on AI in learning and education in Chinese, European, and American contexts. The articles introduce how new intelligent tools and machine learning can support human learning and well-being and what kinds of consequences it has for education and learning environments. The articles provide insights into the state of the art of AI when used in education systems and for learning environments.

The book has four parts:

• (i) AI expanding learning and well-being throughout the life

• (ii) AI in games and simulations

• (iii) AI technologies for education and intelligent tutoring systems

• (iv) AI and ethical challenge in new learning environments

Part I: AI Expanding Learning and Well-Being Throughout Life

The articles cover the methods for how human learning can be supported though AI-based tools and environments in school contexts and informal settings. The articles introduce new methods for how AI-based tools and services can support students’ learning and help them to become more engaged, curious, and in positive social-emotional well-being states. Articles also describe how teachers can be assisted by AI-based tools and environments in diagnosing students’ behavioral and learning difficulties and how researchers can see more deeply into what is happening in classrooms with multimodal data collection.

Part I starts with the chapter “Artificial Intelligence Innovations for Multimodal Learning, Interfaces, and Analytics” of Marcelo Worsley. It describes how the twenty-first century has brought a growing variety of authentic and engaging learning environments. The chapter discusses artificial intelligence-based tools and technologies that can help researchers and practitioners navigate and enact these novel approaches to learning, while also providing a meaningful lens for student reflection and inquiry. The chapter includes technologies that offer insights for using audio/video information and resources for studying learner electrodermal activity, and it provides analytic techniques and interfaces for helping researchers collect and analyze different types of multimodal data across contexts.

Nick Haber underlines in his chapter “Curiosity and Interactive Learning in Artificial Systems” the fact that human learning is interactive, and we learn through curiosity, and we interact with both physical objects and the people around them. This flexible capacity to learn about the world through intrinsically motivated interaction continues throughout life. He asks how we would engineer an artificial, autonomous agent that learns in this way – one that flexibly interacts with its environment, and others within it, in order to learn as humans do. The chapter first motivates this question by describing important advances in artificial intelligence in the last decade, noting ways in which artificial learning within these methods are and are not like human learning. Nick Haber also gives an overview of recent results in artificial intelligence aimed at replicating curiosity-driven interactive learning. Finally, he speculates on how AI that learns in this fashion could be used as fine-grained computational models of human learning.

In the chapter “Assessing and Tracking Students’ Well-Being Through an Automated Scoring System: School Day Well-Being Model”, the research group Xin Tang, Katja Upadyaya, Hiroyuki Toyama, Mika Kasanen, and Katariina Salmela-Aro introduces the model for automated scoring system for modelling students’ well-being. Students’ well-being is critical as it influences their positive development in school life and ensures their future growth. The assessment of well-being has been often static, lagging behind for diagnostic and intervention purposes. In this research, the authors introduce an automated scoring well-being system, School Day Well-Being Model, that is featured as dynamic and real time. User experiences are collected to show the utility of the model. The findings were consistent across the globe.

In the chapter “Learning from Intelligent Social Agents as Social and Intellectual Mirrors”, Bethanie Maples, Roy D. Pea, and David Markowitz introduce the concept of Intelligent Social Agents (ISAs) which are conversational agents that leverage emergent machine learning techniques to present as sufficiently anthropomorphized to pass Turing tests in short exchanges. The interaction capabilities of these agents made possible by advances in artificial intelligence lead to deep emotional bonding with users, leading researchers to reexamine the impact and potential uses of these human-machine relationships in education. In this work, they examined the technical advances that made a new breed of ISA possible, and dive into how one best-in-class ISA, Replika, might be affecting users socially, emotionally, and cognitively. A small, mixed-method study of Replika users explored relationships between user loneliness, use motivations, use patterns, and user outcomes. Their results seem to indicate that the confluence of new functionality, product narrative, and user life stressors make ISAs an emerging tool for cognitive and emotional support, filling a gap in users’ needs which humans do not fill.

Penghe Chen and Yu Lu describe in their chapter “An AI-Powered Teacher Assistant for Student Problem Behavior Diagnosis” a novel interactive technology to diagnose students’ behavioral difficulties in schools. The chapter describes the process of designing and implementing an intelligent teacher assistant, which could advise teachers and help them to diagnose the student problem behavior. Technically, it utilizes a task-oriented dialogue system to help identify the underlying reasons (i.e., the student need deficiency) behind their problem behaviors, and accordingly provides advice to teachers. It also employs the semantic search technology to find the similar cases that have been well resolved by the experienced teachers.

In the chapter “Analysis and Improvement of Classroom Teaching Based on Artificial Intelligence”, Zhong Sun, Zi Chun Yu, and Fei Yun Xu discuss on classroom research and how new AI-based techniques can improve our understanding what happens in classrooms. Common classroom teaching analysis, which focuses on counting and coding teacher-student behaviors and discourse interactions, faces many difficulties as content-free, low efficiency, and small scale in analysis. To overcome the shortcomings of recent research methods, and to foster high-quality classroom teaching, they propose a human and AI technologies blended analysis framework named as TESTII for classroom teaching. It consists of five steps identifying teaching events, sequencing the pedagogies of classroom teaching structure, analyzing teacher-student interaction, interpreting teaching meaning, and providing improvement strategies for high-quality classroom teaching.

Part II: AI in Games and Simulations

This part introduces cross-scientific and multi-method research with cases, pedagogical models for artificial intelligence-supported gaming and simulation-based learning. It starts with an interview of Professor James Lester on narrative-centered learning environments which can be designed as engaging games for students.

In chapter, “Perspectives and Metaphors of Learning: A Commentary on James Lester’s Narrative-Centered AI-Based Environments” is a special chapter by Marianna Vivitsou. It is based on Professor James Lester’s keynote presentation of narrative-centered learning environments. The commentary aims to discuss perspectives on narrative-centered learning and metaphors of AI-based learning. The chapter focuses on the narrative elements that underlies the use of AI in Learning. One example of such environments is Crystal Island, an AI-based game for K-12 students learning science. Vivitsou uses Paul Ricoeur’s narrative and metaphor theories to reflect on the role of characters and the narrative plot in relation to Lester’s visualization of the future of learning with AI-based technologies. In this process, new roles in AI-based learning are introduced. One such example is the role of drama manager. The drama manager is a novel metaphor in game-based learning. In addition, more conventional metaphors, such as the tutorial dialogue, are brought forward as well as technological metaphors. The multiplicity of metaphors have agency at their core. As technological advancement shakes the boundaries of thinking about agency nowadays, new dynamic metaphors are needed in AI-based learning. Toward this direction, the commentary draws from new materialist and post-humanist thinkers to raise these issues and the need to take the narrative further.

In the chapter “Learning Career Knowledge: Can AI Simulation and Machine Learning Improve Career Plans and Educational Expectations?” I-Chien Chen, Lydia Bradford, and Barbara Schneider introduce a game simulation for young adults and those who have lost their jobs. In these life situations, the employment landscape is characterized by ambiguity and insecurity. They introduce the game Init2Winit which integrates data-based analytics with occupational information algorithms that allows users to make choices with respect to their education planning and salary projection in visualizing themselves in a dream job. Their results show promise in terms of the prediction accuracy of educational expectations and users’ behavioral classifications. Init2Winit can be an informational channel for students who lack informal networks in career planning. It also serves as a supplementary network supporting career/ college planning knowledge for students to make better education and employment decisions. Beyond this, the authors propose that machine learning could incorporate a game designed to measure students’ strengths and weaknesses to give career recommendations and pathways.

In the chapter “Learning Clinical Reasoning Through Gaming in Nursing Education: Future Scenarios of Game Metrics and AI”, the research group Jaana-Maija Koivisto, Sara Havola, Henna Mäkinen, and Elina Haavisto introduce how healthcare professionals can improve their clinical reasoning through AI and how AI techniques can be used in healthcare education and training. Previously simulation games have been proven effective for learning clinical reasoning skills. However, game metrics have not been utilized much in nursing simulation games, although research in other disciplines shows that game metrics are suitable for demonstrating learning outcomes. This chapter discusses the possibilities to exploit game metrics in developing adaptive features for nursing simulation games, especially difficulty adoption based on students’ knowledge and skills. Personalization and adaptivity in simulation games can enable meaningful learning experiences and enable nursing students to achieve good CR skills for their future work in constantly challenging clinical situations.

In the chapter “AI-Supported Simulation-Based Learning: Learners’ Emotional Experiences and Self-Regulation in Challenging Situations”, Heli Ruokamo, Marjaana Kangas, Hanna Vuojärvi, Liping Sun, and Pekka Qvist explore learners’ emotional experiences and self-regulation (SRL) and how to overcome stressful situations in a simulation-based learning environment (SBLE). In the experiment, data was collected from the trainees of a basic training phase at Oil Company Neste by online observations, video recordings, and delayed stimulated recall interviews. The findings evidence that SBLE was generally a positive experience to the learners. However, the trainees met several challenging situations with topics related to chemical engineering and process operation. These tasks were often experienced as stressful, and emotional regulation was needed. The trainees used the following SRL operations: metacognitive monitoring, social scaffolding, cognitive operations, and emotional regulation. According to the results, an AI tutor can provide help for decision-making and visualizing critical points of learning processes.

Part III: AI Technologies for Education and Intelligent Tutoring Systems

This part focuses on new systems in which AI technology is used for professional training situated in virtual reality (VR). The articles also describe VR-based learning technology for contextual learning and how scaffolding can be provided by an AI Tutor within VLE. Automatic scoring and e-books are also introduced as tools for improve teaching and learning.

In the chapter “Training Hard Skills in Virtual Reality: Developing a Theoretical Framework for AI-Based Immersive Learning”, the research group Tiina Korhonen, Timo Lindqvist, Joakim Laine, and Kai Hakkarainen develops a theoretical frame for pedagogical settings for an immersive virtual reality-based hard-skills training guided by an artificial intelligence software agent. They suggest the theoretical assumptions of embodied, embedded, enacted, and extended (4E) cognition to fully consider learner epistemology in a virtual world, and to account for and make full use of the unique opportunities afforded by the synthetic nature of the immersive virtual learning environment. They outline a theoretical framework for a virtual reality AI tutor and propose pedagogical principles for such a framework that could inform follow-on research.

The chapter of Shuanghong Jenny Niu, Xiaoqing Li, and Jiutong Luo “Multiple Users’ Experiences of an AI-Aided Educational Platform for Teaching and Learning” provides new knowledge for how AI technology can be used to assist in teaching and learning at schools through The Smart-Learning Partner (SLP) educational platform. This learning environment is based on AI technology to provide new possibilities for individualized learning and more educational resources. The chapter introduces a case study of how the AI-aided SLP platform helped in teaching and learning from students’, teachers’, and a principal’s perspectives at a Chinese school. The platform provided them with diagnostic feedback and assessments, and information about the learning progress. In addition, students had access to various microlectures according to their interests. Teachers got real-time learning reports. They could follow progress at the individual or class level and adjust better their teaching according to students’ needs. The principal used the information in resource allocating and in curriculum planning.

In the chapter “Deep Learning in Automatic Math Word Problem Solvers”, Dongxiang Zhang introduces a new innovative automatic solver for mathematical word problems (MWPs) dated early back to the 1960s. Revolutionary advances of deep learning (DL) have opened new ways to parse the human-readable word problems into machine-understandable logical expressions. The problem is challenging due to the existence of a substantial semantic gap. The chapter introduces various attempts that have been made to bridge the gap, from rule-based pattern matching to semantic parsing with statistical machine learning, and to the recent end-to-end deep learning (DL) models. Despite the great success achieved by applying DL models to solve MWPs, the current status in this research domain still has room for improvement. MWPs have also been recognized as good testbeds to evaluate the intelligence level of agents in terms of natural language understanding and automatic reasoning. The successful solving of MWPs can benefit online tutoring significantly.

The chapter “Recent Advances in Intelligent Textbooks for Better Learning” by Bo Jiang, Meijun Gu, and Ying Du emphasizes that understanding how people read and interact with e-textbooks could not only promote our understanding of how people learn, but also benefit us in providing intelligent learning support to learners. This chapter offers a state-of-the-art overview of intelligent textbooks. It introduces the history of intelligent textbooks and describes the technologies behind these books and what mechanism makes a textbook intelligent. The analysis consists of student modeling approaches from three aspects: the learners’ knowledge state model, the learners’ learning behavior model, and the learners’ psychological characteristic model. The chapter also describes domain modeling technologies. The chapter also summarizes what effects intelligent textbooks provide to students’ learning. The last section discusses the future and challenges of intelligent textbooks.

Part IV: AI and Ethical Challenges in New Learning Environments

This part overviews ethical challenges from Chinese and European perspectives. It also opens up the complex picture of ethical challenges from teachers’ and companies’ perspectives. Games and their algorithms include many ethical questions about transparency and explicability, and these will be reflected upon through a multiplayer game simulation. The part includes also a serious message of risks if AI is used for surveillance.

In the chapter “Ethical Guidelines for Artificial Intelligence-Based Learning: A Transnational Study Between China and Finland”, Ge Wei and Hannele Niemi have reviewed ethical guidelines in China and in Europe where Finland is one member state. The chapter, taking China and Finland as two contextual cases, analyzes how AI-related policies at the national level have focused on educational themes and set aims for improving the quality of learning and education. The references to education are mainly general and indirect, but four themes for AI ethics in education emerged: (1) inclusion and personalization, (2) justice and safety, (3) transparency and responsibility, and (4) autonomy and sustainability. Although both China and Finland recognize the importance of AI ethics, the differences are manifested as policy approaches, properties, and strategies due to sociocultural variation. The authors emphasize the importance of international and transnational dialogue from ethical perspectives to foster our reciprocal understanding of AI and the human-centered stance on education.

In the chapter “Artificial Intelligence Ethics from the Perspective of Educational Technology Companies and Schools”, Päivi Kousa and Hannele Niemi discuss opportunities and challenges that AI is bringing to learning in schools and working life contexts. Ethical issues are viewed from the perspectives of companies who produce educational AI-based tools and services, and from those who use them in schools and workplaces for learning. From companies’ viewpoints, ethical challenges are related to regulations, equality and accessibility, machine learning, and society. From schools’ perspectives, the major critical questions are who has the power to decide which educational services the school can use and who is responsible for the ethical issues of those services, for example, student privacy. In addition, schools are concerned with how to ensure that AI-based services and tools are equally accessible to all and genuinely useful in supporting teaching and learning.

The chapter “Artificial Intelligence in Education as a Rawlsian Massively Multiplayer Game: A Thought Experiment on AI Ethics” by Benjamin Ultan Cowley, Darryl Charles, Gerit Pfuhl, and Anna-Mari Rusanen reflect on the deployment of Artificial Intelligence as a pedagogical and educational instrument, and the challenges that arise to ensure transparency and fairness to staff and students. They apply a Rawlsian justice game, played within the Massively Multiplayer Game: to facilitate transparency and trust of the algorithms involved, without requiring algorithm-specific technical solutions to, for example, “peek inside the black box.” The chapter suggests solutions for the well-known challenges of explainable AI and distributive justice.

The Part IV of ethical issues of AI ends with the chapter “Four Surveillance Technologies Creating Challenges for Education” by Roy D. Pea and doctoral students of Stanford’s Learning Sciences and Technology Design PhD program: Paulina Biernacki, Maxwell Bigman, Kelly Boles, Raquel Coelho, Victoria Docherty, Jorge Garcia, Veronica Lin, Judy Nguyen, Daniel Pimentel, Rose Pozos, Brandon Reynante, Ethan Roy, Emily Southerton, Miroslav Suzara, and Aditya Vishwanath. They summarize four core surveillance technologies that are entering as common practices to universities as well as preK-12 schools: Location Tracking, Facial Identification, Automated Speech Recognition, and Social Media Mining. The authors make several critical questions about how these technologies are shaping human development and learning and how current algorithmic biases increase inequities. They also emphasize that the need for learners’ critical consciousness concerning their data privacy should be taken as a serious task in education. All these challenges need collaboration of government, industry and the public sector.

The final chapter “Reflections on the Contributions and Future Scenarios in AI-Based Learning” by Roy D. Pea, Yu Lu, and Hannele Niemi summarizes the importance of the contribution of all chapters and how they deepen our understanding of what possibilities and challenges exist when AI is applied in education. Seven categories provide perspectives to reflections. Four of them are connected to different levels of the educational system, others are opening scenarios to research on education and learning with AI, and finally the last category is devoted to ethical challenges of AI in education and learning. AI will be the powerful tool in education and learning but ethics of AI in education is a keystone issue which will ramify throughout future inquiries into the future of AI-augmented learning.

5 The Message of the Book

The book is based on interdisciplinary cooperation. Technology and human learning in educational settings are integrated. The book provides examples of the most recent AI research at the nexus of computing sciences, learning sciences, and educational technologies. Much is going on – yet longitudinal studies of emerging and long-term effects are very much needed to understand the dimensions of societal change that education and learning transformed by AI will reveal. The chapters point to the future and give evidence that AI will have significant consequences for education and learning. The book opens up inquiries into how AI supports both students and teachers through interactive, intelligent tutoring, multimodal data and feedback systems incorporating speech, images, and other behavioral data. Many challenges are ethical and related to trustworthy AI and issues of equity in AI applications such as face recognition, games and simulations, personalizing learning, and data mining. It is evident that we will collectively need to continue to develop and report research-based evidence for designing the future toward the benefits of all individuals and their societies.